1. Spectral Detection of Attenuation and Lithology
M. S. Maklad
Signal Estimation Technology Inc.
J. K. Dirstein
Total Depth Pty Ltd.
Good morning Ladies and Gentlemen. It is with great pleasure that I will present “Spectral Detection of Attenuation and Lithology”

2. Overview Introduction
Spectral detection of changes in lithology and stratigraphy
Spectra detection of attenuation
Examples: Canada, Indonesia, Australia
Time-frequency analysis
Conclusions
Here is what I am going to present.
After a brief introduction I will talk about Spectral detection of changes in geomorphology followed by Spectra detection of attenuation.
Then I will present several examples of both techniques from Canada, Indonesia and Australia
Since Time-frequency analysis is currently a topic of considerable interest, I will touch on it briefly.
Finally, I will offer our conclusions.

3. Introduction
Traditional seismic interpretation methods rely heavily on amplitude analysis for quantifying reservoir parameters.
AVO and seismic inversion technologies have been developed to help improve, support and verify seismic interpretation.
Frequency content is another important source if information in seismic data.
Spectral analysis as discussed in this presentation refers to tools used for the examination of the frequency content of times series data.
The presentation will discuss two complementary ways of employing spectral analysis for hydrocarbon reservoir mapping: detection and tuning
Just read the Intro slide!
Traditional seismic interpretation methods rely on imaging where interpreters look for visual structural and stratigraphic features associated with potential reservoirs. This requires the detection, correlation and analysis of seismic features which in turn are better performed on high resolution data. Such techniques rely heavily on amplitude analysis for quantifying reservoir parameters.
Several sophisticated technologies were developed to help improve, support and verify seismic interpretation. The most commonly used are seismic inversion and amplitude versus offset (AVO) analysis . Seismic inversion, in its useful guises as a non-linear deconvolution, attempts to improve amplitude mapping by reducing the effect of interference between close reflectors on their amplitudes. Both inversion and AVO are amplitude-based analysis techniques.
Another important source of information in seismic data is its frequency content. Tools for examining the frequency content of time series data are collectively called spectral analysis. Spectral analysis has established itself as an important tool in seismic data acquisition and processing. In the last two or three decades, applications of spectral analysis to hydrocarbon reservoir detection and characterization emerged [2-5] and have recently intensified [6, 7]
In this paper we explore two complementary ways of employing spectral analysis for hydrocarbon reservoir detection and mapping: attenuation and tuning. We also examine the applications of time-frequency analysis techniques. Finally we present some real data examples.

4. Seismic Spectral Elements
Seismic response
Energy
Source
Earth
Reflectivity
Noise
SEISMIC = + *
Consists of several components:
Wavelet
Reflectivity
Noise
Time
Time Domain
Seismic(t) = Wavelet(t) * Reflectivity(t) + noise(t)
Let us now examine the spectral elements of seismic data;
The wavelet is the propagating entity which travels thru the rocks and is influenced by rock properties.
The reflectivity reflects changes in the acoustic (and elastic) properties of the rocks.
The noise limits our ability to accurately image our zones of interest and the quality and accuracy of the attributes extracted from the seismic.
In the time domain seismic is the convolution of the wavelet with the reflectivity plus additive noise. In the frequency domain we have the seismic frequency response composed of the wavelet times the reflectivity plus noise again.
We strive for a spectral estimate that is robust to noise. We also would like to estimate wavelets for each trace from short time windows. These will be essential elements in the implementation of the technology.
Frequency Domain
Seismic(f) = Wavelet(f) X Reflectivity(f) + noise(f)
Seismic attribute analysis uses information extracted from the seismic data or its constituents.

5. Spectral Signatures Lithological: Petrophysical:
Reflectivity-induced, e.g.
Tuning effects (pinch-outs).
Time thickness changes (sand/shale ratio, porosity, gas).
Faults and Fractures
Petrophysical:
Primarily wavelet-induced (Attenuation)
Attenuation is a function of fluid viscosity, permeability and temperature.
Generally, gas attenuates more than oil and oil more than water.
Reflectivity may be affected by attenuation as well.
we now look at what affects the spectral elements of seismic data.
Lithologically,
the spectral character changes in response to time-thickness variation within a formation or a group of formations....usually, these variations can be associated with changes in velocity within the formation. Petrophysically
Spectral attributes can be used to estimate the attenuation characteristics of a certain formation. It has been experimentally established that fluid-bearing porous rock formations attenuate seismic waves preferentially, i.e. higher frequencies within the seismic band , are more severely attenuated than lower frequencies. Generally, gas attenuates more than oil and oil attenuates more than water.

6. Spectral Detection of Lithology
Short Analysis Window
waveform pattern 1
waveform pattern 2
waveform pattern 3
Time
This schematic shows three different lithologies all encased in a common environment. The seismic signature of the three lithologies may not differ appreciably. Going to the frequency domain may produce significantly different spectral patterns.
Many stratigraphic features such as changes in sand/shale ratio may induce small changes in the time-thickness which could translate into more detectable changes in the frequency domain. Another example would be the detection of reef edges where there would be a dramatic change in the spectral character of the data. Backscattering from faults may also affect the spectral character of the data near faults. Anisotropic changes in velocities will be associated with changes in the time-thickness of a formation from one azimuth to another which could be easily detected spectrally.
Here it is crucial to be able to reliably estimate the spectrum from as short a window as possible so that changes in the zone of interest would be the dominant factor in the spectra.
spectral pattern 1
spectral pattern 2
spectral pattern 3
Often, spectral patterns are more distinct than waveform patterns.

7. Devonian Pinnacle Reef Example:
Our first example is an illustration of lithologic detection on a real data example of an Upper Devonian pinnacle reef from Western Canada.
The uppermost panel shows a seismic line tying three wells which have been drilled in search of pinnacle reefs. The dry well on the right failed to encounter carbonate reservoir. The oil well intersected 50 m of oil pay. The dry well to the left encountered some carbonate and oil shows but no movable hydrocarbons. The carbonate reservoir is a stratigraphic trap encased in sealing shales. Seismically, the reef lies in the middle of the analysis window which is shown in red. With this analysis window size and position which includes the reservoir, we expect to look at the lithological effects on the spectral content of the data.
The middle panel contains the estimated wavelet spectra of the seismic data. Note the shift to lower frequencies and the narrowing of the bandwidth over the reef. the lowermost panel shows the residual wavelet spectra after removing the common component. This has enhanced the spectral anomalies allowing for a more accurate prediction of the reef edges and perhaps, an untested second pinnacle reef.

8. Attenuation Wb(f) Wa(f) Attenuation = Reservoir Wavelet above -
Wavelet below -
|Wb(f)| Wa Wb
Reservoir
ZOI
Time
Now we shall discuss how attenuation could be estimated from seismic data. The wavelets reflected from interfaces above and close to the top of the reservoir are a good representation of the wavelets that will start propagating through the reservoir. This is annotated on the slide as Wa(f).
The wavelets that reflected from below and close to the base of the reservoir are a good representation of the wavelet that completed propagation through the reservoir. This is denoted on the slide as Wb(f) . We assume that the major source of change in the spectral character of the wavelets is the reservoir itself (i.e. wavelets above the reservoir are well approximated by one wavelet and the wavelets below the reservoir are well approximated by a second wavelet).
Attenuation is then calculated from the two different estimates of the wavelets.
To obtain a reliable indication of attenuation due to pore fluid we need to preserve spectral integrity.
For example: No time varying filtering, Minimization of coherent noise such as Multiples and converted waves.
Etc.,

9. 4D Monitoring: Cold Lake
Dilay, A J, Maklad, M S and Eastwood, J., “Spectral Analysis Applied to Seismic Monitoring of Thermal recovery,” Extended Abstracts of the 1993 SEG International Meeting, Washington, pp
1.0
0.8
0.6 50
Production
Injection
0.4
0.2
0.0
100
150
200
Frequency 1
Power (normalized)
Above
Our first example for attenuation is a revisit to a presentation written by Dilay, Maklad and Eastwood and presented at the 1993 SEG conference.
The example shows the application of spectral analysis to seismic data acquired over the Cold Lake Bitumen deposit located in eastern Alberta.
The main purpose of this 4D is to monitor the reservoir between injection and production phases.
During the injection cycle steam at high temperature and pressure is injected at specified locations. As the Bitumen is heated it produces gases at the same time the steam loses temperature and pressure and becomes more gaseous. Therefore, the heated zone of the reservoir would exhibit significant attenuation compared to the unheated zones. Thus, attenuation can help map the steam front propagation limits which is essential in planning production well locations and the next phase of the injection cycle.
The power spectra on the upper right shows the spectral response above the reservoir for two separate cycles. Here we can observe good spectral repeatability between the two surveys. The lower graph shows the spectra from below the reservoir at the same location. Notice that there is significant spectral attenuation during the production cycle.
Power Spectra above the reservoir shows good spectral repeatability between the two surveys.
Below the reservoir higher frequency attenuation is evident in the production cycle.
Below

10. 4D Monitoring: Cold Lake (85% quantile frequencies)
Injection
High Temp & Fluid Press
Low Gas Saturation
Production
Low Temp & Fluid Press
High Gas Saturation
Above
Power spectra for time window above the reservoir show high repeatability for production and injection surveys
High frequency attenuated zones during oil production correlate spatially with the 15 CSS wells for spectral window below reservoir
High frequency attenuated zones correlate with velocity sag anomalies.
Most probable cause of high frequency attenuation is gas saturation in reservoir
The upper maps show the 85th quantile energy surface above the reservoir for the injection (left) and production (right) cycles. Color scale is from Hz (red to white). Notice there is a good degree of similarity.
The lower maps show the 85th quantile energy surface below the reservoir for the injection (left) and production (right) cycles. Color scale is from Hz (red to white). Spectral differences between injection and production cycles are very pronounced, in some locations as much as 30 Hz . They show strong attenuation of high frequencies during the production cycle, which the authors attribute to changes in partial gas saturation.
Upper Left
Spectral tracking for window above reservoir from 1992 injection survey. This slide show the 85% quantile energy surface for the above-reservoir window for steam injection plotted in map view. Color scale is from Hz (red to white). Compare with slide 15.
Upper Right
Spectral tracking for window above reservoir from 1990 production survey. This slide shows the 85% quantile energy surface for the above-reservoir window for oil production plotted in map view. Color scale is from Hz.
The spectral character across the whole 3-D survey described by these high-frequency energy surfaces is very similar during both injection and production, again attesting to good spectral repeatability above the reservoir.
Lower Left
Spectral tracking for window below reservoir from 1992 injection survey. This slide show the 85% quantile energy surface for the below-reservoir window for steam injection plotted in map view. Color scale is from Hz, red and pink delineating zones of strong attenuation of high frequencies. Compare with slide 17.
Lower Right image
Spectral tracking for window below reservoir from 1992 production survey. This slide shows the 85% quantile energy surface for the below-reservoir window for oil production plotted in map view. Color scale is from Hz.
Spectral differences between injection and production cycles are very pronounced, as much as 30 Hz in some locations. They show strong attenuation of high frequencies during the production cycle, which we attribute to partial gas saturation. Furthermore these attenuation anomalies correlate spatially with the locations of the 15 wells on the pad, consistent with known pressure gradient and fracture communication pathway.
Below

11. Tubridgi: Conventional Gas
Spectral Signatures of the Tubridgi Field: Onshore Carnarvon Basin, Western Australia
J.K.Dirstein & M.S. Maklad PESA News: 1997 & 2007
The Tubridgi Gas Field is located 30 Km west-southwest of Onslow W.A. in the onshore portion of the Carnarvon Basin in Production License L9. The hydrocarbons are entrapped in a northeast trending anticlinal structure with broad, low relief evident in depth mapping of seismic two-way-times. The Early Cretaceous and Mungaroo reservoirs are sealed by the Cretaceous Muderong shale.
In May and June 1997, Total Depth Pty Ltd undertook a study using proprietary technology developed by Signal Estimation Technology Inc. to extract and analyse spectral attributes from selected seismic lines across and adjacent to the Tubridgi Field. The objectives of the study were as follows:
To determine whether spectral attenuation signatures could be measured beneath the Tubridgi Gas Field and used to discriminate between gas wells and dry wells.
To determine the distribution of the observed spectral attenuation signatures and determine their use with regard to the future development of the Tubridgi Field and remaining exploration potential of the surrounding area.
Currently, thirteen seismic lines from three different vintages have been analysed, in two stages, and spectral results have been calibrated using ties with sixteen wells.
The Tubridgi Gas Field is located 30 Km west-southwest of Onslow W.A. in the onshore portion of the Carnarvon Basin.
The hydrocarbons are entrapped in a northeast trending anticlinal structure with broad, low relief evident only in depth mapping of seismic two-way-times. The Early Cretaceous and Mungaroo reservoirs are sealed by the Cretaceous Muderong shale. .

12. Tubridgi: Spectral Attenuation Validation
Spectral Attenuation panels at 16 Boreholes
Image on the left is a composite showing panels of spectral attenuation (color) and wavelet spectra (grey), taken from seismic data ties with 16 boreholes. Note that there is a greater than 90% correlation with the presence of hydrocarbons and the higher frequency seismic attenuation. Follow-up drilling of six wells in the field over a period of several years continued to demonstrate the relationship between high frequency attenuation and the presence of hydrocarbons.

13. Perseus

14. Perseus Gas Field: Offshore NW Australia
1997 Mobil Oil WA-248P Attribute Analysis
Zone of
Interest
Attenuation Spectra:Black = Maximum Attenuation
Stratigraphic Gas Field
Frequency
TWT
Discovered in 1996, the Perseus gas field is about 135 km northwest of Karratha in 131 m of water and started production in The Athena fi eld was discovered in October 1997 and is an extension of the North West Shelf Gas project’s Perseus gas fi eld. The Athena-1 well was drilled in a water depth of 120 m and reached a total depth of 3364 m. The well was tested over four zones and achieved a combined fl ow rate of 1340 kcm/d (47.4 MMcf/d) of gas and 2133 bbl/d of condensate. A production licence over the Athena fi eld was awarded in January 1999.In early March 2001, Mobil Australia Resources Company Pty Ltd and Phillips Australia Gas Holdings Pty Ltd signed an agreement with the North West Shelf Venture participants in relation to the development of the Perseus–Athena gas fi eld. Under the agreement, Woodside Energy Ltd. as operator of the North West Shelf Venture, is producing gas from the WA-17-L permit on behalf of the permit holders. Production is through the North Rankin A production facility and the term of the contract is for the life of the Perseus fi eld.The Athena fi eld commenced production in late 2001.
Discovered in 1996, the Perseus gas field was drilled in 131 m of water and started production in The Athena field was discovered in October 1997 and is an extension of the North West Shelf Gas project’s Perseus gas field.

15. Badak Field: East KalimatanNW
Time
Zone of
Interest
The Badak field was discovered in 1972 and a seismic line templating the field is shown. Hydrocarbons are trapped in a large anticlinal structure in stacked accumulations at depths from to 3600 metres. The spectral example shown in this study examines a zone of productive Upper Miocene sandstone reservoir around 1700 metres below the surface.
The Badak Oil and Gas Field is located in the onshore portion of the Kutei Basin in the northern portion of the Mahakam delta system of eastern Kalimantan.The Badak field was discovered in 1972 and a seismic line templating the field is shown in Figure 2. Hydrocarbons are trapped in a large anticlinal structure in stacked accumulations at depths from to 3600 metres. The spectral examples shown in this study examine a zone of productive Upper Miocene sands tone reservoir around 1700 metres below the surface.
1 Km
Seismic Line K7802 East Kalimantan
Spectral Signatures of the Badak Oil and Gas Field: Onshore Kutei Basin, Kalimantan, Indonesia: Dirstein, Maklad et. al IPA conference.

16. Badak Field: Spectral Attenuation
TWT
Frequency
Attenuation
Estimated Wavelet Spectra
Percentile
Frequencies
Badak Oil & Gas Field
C08
Analysis
Window
Attenuation Spectra
The attenuation spectra shows an attenuation anomaly between Hz.
The presence of hydrocarbons in the Miocene Sandstone reservoirs at the Badak field appears to have caused measurable attenuation of higher seismic frequencies.
The attenuation measurement was made from geophysical archived stacked seismic data.

17. Spring Grove: Surat Basin Qld Australia
“A Major Play In The Surat – Bowen Basin”
Talk presented by Dr Howard Brady, CEO of Mosaic Oil NL, to PESA NSW Tuesday March 9th 2004
D&A
Target reservoir is a Permian aged sandstone reservoir. Initial well drilled crestally on structure failed to encounter reservoir sand at target level (D&A).
Play concept: better sand development off paleo-high. Which flank would be the better location to test this play concept?
Spectral attenuation showed an anomaly on the eastern flank of the structure. Subsequent well location suggested by this anomaly was an oil well (no associated gas). The field has expected recoverable reserves of around one million barrels.

18. Time Frequency Analysis
Provides better time localization of the frequency content of seismic data
First Technique ultilized STFT which is basically a sliding window FT. Resolution depends on the length of the window. Both low and high frequencies would have same analysis bandwidth (i.e. higher frequencies more accurately resolved than lower frequency).
The Gabor transform is one of the earliest implementations of the STFT with a Gaussian window
Wigner-Ville distribution provides high resolution time-frequency analysis employing a quadratic transform obtained by instantaneous correlation of the signal. This approach suffers from interference resulting from its quadratic nature.
Note that geology and wavelet effects are not separated in time frequency distributions.
Better localisation in time of the frequency content of the data compared to Fourier analysis. A sine wave is localised in frequency but infinite in time, whereas a spike is localised in time and infinite in frequency.
Two signals can have the same amplitude spectrum but different time distributions of their frequency content. Time frequency analysis offers a powerful tool to discriminate between such signals.
Short-Time Fourier Transform (STFT)

19. The Wigner-Ville Distribution: Cross Term
A signal composed of two modulated Gaussians (Gabor atoms)
In this example the signal is composed of two distinct oscillations; an earlier low frequency pulse and a later higher frequency pulse.
The lower graph shows the time-frequency distribution of this signal computed using a Wigner-Ville distribution. We see a low frequency area around the centre of the first pulse and another higher frequency around the centre of the second pulse. So far, So good. However, we also have something in between that we cannot interpret in light of what we know. This is the cross-term which limits the usefulness of the Wigner-Ville technique.
Decomposing the signal into time-frequency wavelets addresses the cross-term problem.
The time-frequency distribution of the signal is obtained by adding the Wigner-Ville distribution of the individual components.
Note that different methods for time-frequency analysis lead to different results.
For Example: STFT would not show the cross-term, but would smear the spectra
Since the decomposition can be undertaken with different wavelets, this would lead to different time-frequency results.
(i.e. Inversion with the wrong wavelet leads to dispersion of each reflection coefficient into several coefficients and a loss in resolution).
WV distribution of above signal. Notice the interference term.
Decomposing the signal into time-frequency wavelets addresses the cross-term problem.
The time-frequency distribution of the signal is obtained by adding the Wigner-Ville distribution of the individual components.

20. Time Frequency Analysis: Applications
1. Detection of changes in lithology.
2. Detection of coherent noise.
3. Crude attenuation indicator.
Wedge Model
Time-frequency analysis has many potential applications including: detection of lithology, detection of coherent noise such as multiples and converted waves, detection of anisotropy and as a crude attenuation indicator.
The images on this slide show time frequency analysis using sparse decomposition on two synthetic wedge models. The second model contains a shale plug. The display for both examples shows the 16hz amplitudes. The shale plug is delineated by the high amplitude within the wedge. Otherwise the two responses are identical. Note that red indicates higher amplitudes.
Note the changes in reflectivity are identified very well by the time-frequency analysis, however, these changes are clearly a result of changing reflectivity and not an indication of attenuation.
Wedge Model with shale plug

21. Conclusions:
Target-oriented spectral attribute analysis can provide an effective indicator of change within the zone of interest.
Depending on the analysis method, the results can provide and indicator of changes in geomorphology or pore-fluid.
Demonstrated several examples of spectral attenuation in both oil and/or gas reservoirs.
These spectral attenuation signatures have been observed and documented by the authors and collaborators over many oil and gas accumulations on 2D/3D/4D seismic data worldwide at depths from 200 to 3000 meters in Cenozoic, Mesozoic and Paleozoic aged section in clastic, carbonate and fractured reservoirs.
Time-Frequency analysis is more indicative of changes in geomorphology than pore-fluid content.

22. Powerful Scientific Tools for all Phases of the Life Cycle of your Assets